EE234 V 01

Engineer-to-Engineer Note
EE-234
Technical notes on using Analog Devices DSPs, processors and development tools
Contact our technical support at dsp.support@analog.com and at dsptools.support@analog.com
Or visit our on-line resources http://www.analog.com/ee-notes and http://www.analog.com/processors
Interfacing T1/E1 Transceivers/Framers to Blackfin Processors via the

Serial Port
Contributed by K. Unterkofler and T. Lukasiak
Introduction
This EE-Note describes how to interface
Blackfin processors to standard T1 or E1
encoded signals. The proposed template assumes
that a Blackfin processor is functionally located
between two T1/E1 streams to perform the
required processing, such as line echo canceling.
However, the Blackfin processor can be adapted
easily to other functional architectures. As shown
in this EE-Note, most standard backplane PCM
data streams interface directly to the processors
serial port(s), without any external hardware.
This document provides schematics, layout
suggestions, and a software framework for
receiving, processing, and transmitting PCM
streams between two T1/E1 transceivers/
framers.
The chosen framing device is the PMC-Sierra
PM4351 COMET. The PCM streams are
connected to one of the processors synchronous
serial ports (SPORTs), which can handle two
input streams and two output streams. The
COMET is configured via the processors
asynchronous memory interface in the external
bus interface unit (EBIU).
The schematics are intended to be for a daughterboard that plugs into an EZ-KIT Lite
evaluation system available from Analog
Devices, Inc.
The application was implemented on and is
described herein for the ADSP-BF533 processor
Rev 1 May 11, 2004
[1], [2]. Though it is possible to apply this same

board to the ADSP-BF561 EZ-KIT Lite
evaluation system with only minor changes,
considering that this device incorporates two
cores and two SPORTs, it makes more sense to
extend the board to four COMET devices and to
allow it to handle twice the number of processed
channels [3].
The board was also designed to interface

also to a ADSP-BF535 processor,
although it requires more substantial
hardware changes, such as using two
SPORTs instead of the one needed for the
ADSP-BF533. The schematics refer to the
the required changes. This entails
significant software which are beyond the
scope of this EE-Note.
The software framework is written in C entirely.

All references to execution times are taken from
this code example.
System Architecture
Figure 1 shows a block diagram of a typical
application. The board is designed to plug into an
existing T1/E1 connection via two RJ48C
connectors. This leads to two streams (stream 1
and stream 2), both carrying incoming and
outgoing data (RIN, ROUT, SIN, and SOUT). In the
most basic mode of operation (pass-through),
RIN and SIN are copied unaltered to ROUT and
SOUT, respectively, such that the whole resembles
the original single T1/E1 connection.
Copyright 2004, Analog Devices, Inc. All rights reserved. Analog Devices assumes no responsibility for customer product design or the use or application of
customers products or for any infringements of patents or rights of others which may result from Analog Devices assistance. All trademarks and logos are property
of their respective holders. Information furnished by Analog Devices applications and development tools engineers is believed to be accurate and reliable, however
no responsibility is assumed by Analog Devices regarding technical accuracy and topicality of the content provided in Analog Devices Engineer-to-Engineer Notes.
a
T1/E1 Interface Board
T1/E1 Stream 1
T1/E1 Stream 2
Rout
Rin
Framing and
processing
Sout
Sin
RS232 to Host
Figure 1. System overview
The software also provides a parser-based UART

interface to a generic host processor for setting
operating modes and parameters and controlling
the Blackfin processor and the COMET framing
device.
Figure 2 provides a detailed view of the interface.
The signals from the two RJ48C connectors are
conditioned and isolated from the digital domain
by transformers. Also provided are over-voltage
protection diodes. The COMET framers
(PM4351) are configured to generate serial inputs
in a format compliant with the SPORTs multichannel mode from the RIN and SIN streams, and
to generate the ROUT and SOUT streams from the
SPORTs serial output data. The two COMET
framer chips are connected to the Blackfin
processor via the External Bus Interface Unit
(EBIU) and are configured via this interface.
T1/E1 Data Formats

T1/E1 connections are a well-known and widely
adopted standard. For the purpose of this EENote, it is sufficient to say that T1 lines encode 24
channels of 8-bit data plus one framing bit

(totaling 193 bits) into a 1.544 MHz carrier wave.
Similarly, E1 lines encode 32 channels of 8-bit
data (totaling 256 bits) into a 2.048 MHz carrier.
In both cases, this corresponds to a new sample of
each channel every 125 s. The 8-bit data is
obtained from linear, 14- or 13-bit data samples
by compression according to a-law or -law,
respectively.
In addition to the data (sometimes referred to as
the payload), T1/E1 streams contain signaling bits
according to a variety of communication
standards. This EE-Note does not go into the
details of these standards, since the COMET
devices functionality is to extract the payload
from the incoming T1/E1 streams and to encode a
valid T1/E1 outgoing stream from the payload
from the DSP, according to the selected standard.
Thus, if the COMET devices are set up correctly,
the processor will see the payload only and
never have to handle the signaling bits. If errors
occur, the COMET can be set up to generate an
interrupt for the DSP as well. For more details,
refer to [4], [5] and [7].
Interfacing T1/E1 Transceivers/Framers to Blackfin Processors via the Serial Port (EE-234)
Page 2 of 13
Sout
CLK
2.048MHz
RS232
Protection
Line Interface
Protection
&
&
UART
DSP
Isolation
Isolation
EZ-KIT Lite
via
Expansion
Connectors
COMET1
RJ48C
Line Interface
RJ48C
Rin
Rout
Sin
COMET2
EBIU
SPORT
PM4351
PM4351
PCM data streams
Figure 2. T1/E1 Interface Card Block Diagram
The incoming streams (RIN and SIN) are thus

converted (mapped) into a PCM data stream,
sometimes referred to as backplane. Again,
there are a variety of standards, but they consist of
a signal containing the data bits for each channel
and, for T1 lines, the framing bit, synchronous to a
1.544 (T1) or 2.048 MHz (E1) clock. This block
of data is called a frame. The beginning of a frame
is signaled by a separate signal called a frame
sync. For T1, the COMET also provides the
option of mapping the 24 channels into a
2.048 MHz backplane, such that channels 24-31
contain dummy data, which the processor can then
ignore. This offers the advantage that both T1 and
E1 settings use the same clock, thus switching
between the two formats requires only a simple
software change, rather than switching between
two clock sources.
For the outgoing streams (ROUT and SOUT), similar
considerations apply. The backplane signals
consist of 32 channels of 8-bit data for an E1 line.
Similarly, for a T1 line, the COMET simply
ignores channels 24-31.
Figure 3 graphically shows the backplane formats.

Since each SPORT has two transmit lines and two
receive lines, the ADSP-BF561 can be used to
handle all four PCM streams. The clock and frame
sync are shared so that the data flows are
synchronized at all times. With its ability to
independently select driving and sampling edges
for the receiver and the transmitter, clock and
frame sync, and delay between frame sync and
first bit of the data streams, the ADSP-BF561 can
be adapted to virtually all common backplane
formats. Figure 3 shows the format used in this
application.
For the PCM data to be available for processing, it
must be transferred into the processors memory.
Similarly, outgoing data has to be transferred from
memory to the SPORT. This is done via direct
memory access (DMA) channels. The so called
2D-DMA capabilities allow for rearranging
samples in flexible ways, as is shown in
Figure 4. Incoming samples are stored in the
manner depicted in the figure as they arrive, that is

RIN[0],
SIN[0],
RIN[1],
SIN[1],
Page 3 of 13
a
RIN[Number_of_channels-1], SIN[Number_of_channels-1]
for frame 0, and similarly for the other frames.
Note that the two receive lines (RIN and SIN) are
re-organized as they come in to the SPORT
without any intervention of the core. The attached
code
example
acquires
a
block
of
Number_of_Samples frames, set by default to 40,
such
that
an
interrupt
occurs
every
(Number_of_Samples * 125) s (5 ms for the
provided example). The interrupt signals that the
acquired frames are available for processing, but
the DMA continues to acquire new frames in a
different memory area. When this second block is
acquired, another interrupt is generated and the
DMA places new frames into the first memory
area, overwriting the ones already there. This
mechanism is sometimes referred to as double
buffering and allows the Blackfin processor to
process half the incoming data (the first
Number_of_Samples frames) while the other half is
acquired. This is done in hardware, without any

code intervention, by using the descriptor chain
mechanism provided by the DMA engines.
Similar considerations apply for the transmitted
streams, ROUT and SOUT. For details on DMA
functionality, refer to [2].
The last step before processing the incoming data
is to convert the PCM data into linear values. As
previously mentioned, the incoming data is
compressed and has to be expanded using a-law or
-law. Once processed, the resulting output data
must be compressed to fit into the outgoing PCM
data streams.
The SPORT does the expanding and compressing
(companding) in hardware, which eliminates
processing time needed by the processor.
Figure 3. Backplane signal formats
Page 4 of 13
a
Ordering represents
increasing 16-bit
word addresses
Frame
[0]
Frame
[1]
Rin[0]
Sin[0]
Rin[0]
Sin[0]
Rin[1]
Sin[1]
Rin[1]
Sin[1]
...
...
...
...
Rin[Number_of_channels-1]
Sin[Number_of_channels-1]
Rin[0]
Sin[0]
Rin[0]
Sin[0]
Rin[1]
Sin[1]
Rin[1]
Sin[1]
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
Rin[0]
Sin[0]
Rin[0]
Sin[0]
Sin[1]
Rin[1]
Sin[1]
...
...
...
...
Frame
Rin[1]
[Number_of_Samples-1]
Repeat from
beginning
1st Half
2nd Half
Figure 4. PCM samples allocation in L1 Data Memory see text on how 2D-DMA fills the buffers
Note that the hardware companding does not

guarantee bit accuracy; thus, if an incoming
sample is expanded and then compressed again,
the result may differ from the original by one bit.
This may not be acceptable, depending on the
application. By default, the provided examples use
software companding, which provides bit
accuracy, but requires about 75 s or 65 s of
processing time, each way, for a-law or -law,
respectively, measured for 40 frames at a core
clock speed of 600 MHz. To put this in
perspective, this is about 3.0% (or 2.6%) of the
overall available computation time of 5 ms.
The
described
double-buffer
mechanism
introduces an intrinsic delay. For instance,
considering a single data word that enters RIN, it
will be copied to ROUT in the next block of
frames, and then transmitted with the block after
that. Thus, the delay from RIN to ROUT is (2 *
Number_of_Samples * 125) s (10 ms for the
provided example, but 250 s in the best case).
Similarly, samples going into SIN will reappear

(unaltered or in processed form) on SOUT with the
same delay. The minimum amount of delay is
clearly dependent on how many frames the
adopted processing algorithm requires at each
iteration to produce the desired results.
Reference Board Design

The design closely follows the reference design
described in great detail in [6]. In particular, the
line interface with its signal conditioning, overvoltage protection, and isolation has not been
modified. The analog side of the COMET devices
is connected to the line interface in accordance
with the reference design as well. The 2.048 MHz
clock is derived from an external oscillator device,
which also feeds the processors SPORT clock.
The frame sync is generated from the Blackfin
processor. Since both COMET devices have the
same external clocks and frame syncs, they are
Page 5 of 13
a
forced to synchronize all four PCM streams. They
come from, and go to, the SPORTs primary and
secondary transmitter, and primary and secondary
receiver, respectively.
1. Check whether a message has been received

from the UART host, and take the appropriate
action.
The COMET devices are configured via a 9-bit

address bus and an 8-bit data bus. These buses are
mapped onto the EBIUs external buses. The
devices are selected by the /AMS3 and /AMS1
memory bank select signals.
2. Check whether a new block of received data is

ready for processing and a new block of
outgoing data is empty to hold new results
(via two semaphores set by the DMA interrupt
service routines). If it is not, repeat the cycle
above, otherwise go to step 3.
3. Process the data according to the content of

global
and
per-channel
configuration
variables. Three modes are available:
The address decoding scheme adopted in

this design is very simple. It can be
improved by using one memory bank
select signal only, and decoding more
address lines. This requires one or two
logic gates.
Be careful when laying out the EBIU
connection to the COMET devices.
Although the interface is that of a slow
asynchronous memory device, the
EBIUs buses are shared with SDRAM
devices, which can operate at up to
133 MHz. If SDRAM accesses are
required (for example, when complex
algorithms must be executed), layout of
these buses becomes critical.
The reference design, which is configured as a

daughter-board to the EZ-KIT Lite board, buffers
the EBIU buses and uses a flag pin (PF4) to
optionally disable the buffers outputs when
access to the COMET devices is not needed.
Software Framework
Figure 5 shows a high-level flowchart of the
provided framework. Because the code is modular
and well commented, this EE-Note provides only
a brief outline. More importantly, this section
gives an overview on the possible operating
modes and options that are implemented.
After the initialization of the hardware (Blackfin
processors PLL, SDRAM, asynchronous memory
controllers, SPORT, DMA and exception handler;
COMET devices; and the UART interface), the
main loop performs three basic tasks:
a. Global pass-through: for initial debugging,

all data is passed through unaltered, and, in
case of software companding, not even
expanded and re-compressed.
b. Global Enable, Per-channel Enable for
current channel is OFF: data for that
particular channel is passed through
unaltered; however, unlike the previous
case, the incoming data is expanded,
copied to the output buffer, and then recompressed. Since it works on a perchannel basis, this mode is intended to be
used for individual pure-data channels, as
opposed to voice channels.
c. Global Enable, Per-channel Enable for
current channel is ON: data for that
particular channel (RIN and SIN) is fed as
input to the desired algorithm (such as a
line echo canceller) and the algorithm
places its results into SOUT. In addition,
each channel can also be scaled
individually before entering the algorithm
(separate scaling factors for RIN and SIN)
In all cases, ROUT is a copy of RIN. You can
switch between these three modes at run-time
by setting the corresponding variables from
the UART host. The scaling factors and the
companding law can be changed at runtime
via this interface as well.
Although the above modes and settings are
dynamic, many settings must be determined at
Page 6 of 13
a
compile time. All of them are controlled by
(un)commenting or assigning values to #define
clauses in one file system.h. The most important
ones are:
!
CLKIN, CORECLK, SYSCLK, and SPORT0CLK:

set system frequencies (in Hz)
Number_of_Samples: number of frames acquired
Settings for backplane clock source, frame

sync source, and driving and sampling edge
relationships for COMET and SPORT. Only
needed if adapting to a different backplane
timing standard than what is shown in Figure
3.
Assignments of General-Purpose Flags for

various functions. Flag 0 and Flag 2 are used
to reset the COMET devices and generate
interrupt requests, respectively.
DEVICE1, DEVICE2: Base addresses in external
for each block

!
Number_of_Channels: Set to 32 for a 2.048 MHz
backplane; set to 24 for a true T1 (1.544 MHz)

backplane
!
asynchronous memory of the two COMET

devices. By default, they are located at the
start of bank 3 and bank 1, respectively
SOFTWARE_COMPANDING: if defined, enabled
companding in software, otherwise the

SPORTs hardware companding feature is
enabled
!
COMPANDING_LAW
T1_map: defines settings for T1 (mapping 24
(available
only
if
hardware companding is selected): sets the
SPORTs companding law
STANDALONE: if this label is commented, the

application writes diagnostic and status
information to the VisualDSP++ Output
window.
DUMP_VARIABLES: sends a message with
addresses of important mode variables over

the UART terminal, used to control operating
modes at runtime.
T1 channels onto a 2.048 MHz backplane)

!
E1, T1: define settings for an E1 backplane @

2.048 MHz or a T1 line @ 1.544 MHz,
respectively
E1 and T1 COMET configuration

settings are not implemented at this
time. If these configurations are
required, modify the
initializeCOMET_T1_Short_Haul()
DEBUG_COMET: verifies every write to the

COMET registers and displays a warning if
the read value does not match the originally
intended value. Used for debugging.
DEBUG_ALGORITHM: uses flag pins to show
execution time of the algorithm, etc. Used for

debugging, see Flag Settings in system.h
or
initializeCOMET_E1_75()
functions in COMET_init.c according to

the procedures in [4] and [7].
Page 7 of 13
a
Main()
ProcessData()
Set Exception Handler

Set System Frequencies
SDRAM controller
ASYNC memory controller
Flag IO
Reset COMET devices
Initialize variables
Calculate which half of

databuffers can be
processed
Detect UART terminal

and set up connection
Press (or send)

@ on Terminal
for AutoBaud
detection
Copy Rin to Rout

(Number_of_Samples *
Number_of_channels)
words
Initialize COMET devices

PASS
through?
Initialize SPORT and DMA
Initialize Interrupts
No
Copy SIN to SOUT
Number_of_channels)
words
Expand Rin and Sin

with a-law or u-law
Start Transfers
Loop: For EACH channel

(Number_of_Channel times) :
Process Loop
(Infinite)
Voice
Channel?
Message from
Host?
Yes
Yes
Input Scaling Rin and Sin

Process the
message
Semaphore:
new Block of
Samples?
Semaphore is set by
DMA interrupt every
Number_of_samples * 125usec
Sout() =
Algorithm(Rin,
Sin)
Copy expanded
Sin to Sout
Number_of_channels)
words
Output Scaling Sout
Yes
Clear Semaphore
ProcessData()
Compress Sout
with a-law or u-law
End Channel Loop
return
End Process Loop
Figure 5. Flow Diagram of Main() routine and Processing of PCM samples
Page 8 of 13
a
Memory Allocation Data Cache
The software framework has been structured
knowing that most algorithms will work with
relatively small amounts of code but operate on
large amounts of data. Therefore, the code is
assumed to be small enough to fit into L1
instruction memory thus, executing at the
processors core clock frequency. The software
framework enables data caching for eventual large
data buffers in external SDRAM. This is simply
done by the initial definition of the int __cplb_ctrl
control variable in main.c. The data cache
organization is done in the cplbtab533.s file and
excludes external asynchronous memory from
being cached (such that there are no caching
effects on the COMETs configuration and status
registers).
The data buffers for the incoming and outgoing
PCM data streams are placed into (non-cached)
L1 memory since they are modified by DMA
transfers, which are independent of the cache
system.
In conclusion, the framework places instructions
and all necessary data into non-cacheable L1
memory and sets up the upper half of L1 Data
Bank A as cache buffer for data placed in
SDRAM. This approach has been tested on a line
Command Syntax
echo canceller algorithm, where caching reduced

the execution time for each channel from roughly
1 ms to 100 sec.
Using the UART Host Port

The UART port is used to interface with a host
processor using easy-to-read string commands.
This allows the host to be an embedded
microcontroller or a PC with a human operator.
An interrupt-driven system is used to receive data
from the UART. Each received character is stored
in a temporary array and echoed back to the
UART as an acknowledge signal. The host must
prevent buffer overruns based on this
acknowledgment. As soon as the CR character
(carriage return) is received, the temporary char
array is placed on a queue of strings to be parsed.
Table 1 lists a few of the general commands
implemented in the prototype system. The
READ_MEM and WRITE_MEM commands modify
the processors memory in general, and the control
and status variables in particular.
To create a system where new commands can
easily be added, two tools (Lex and Yacc) were
used to make sense of the incoming string queue.
Purpose
COMET_READ_REG <base addr> <reg addr>
Read a register from the COMET framer
COMET_READ_REG <base addr> ALL
Read all registers from the COMET framer
COMET_WRITE_REG <base addr> <reg addr> <val>
Write a register to the COMET framer
READ_MEM_8 <addr>
Read an 8-bit value from a location in Blackfin memory
READ_MEM_16 <addr>
Read a 16-bit value from a location in Blackfin memory
READ_MEM_32 <addr>
Read a 32-bit value from a location in Blackfin memory
WRITE_MEM_8 <addr> <val>
Write an 8-bit value to a location in Blackfin memory
Write a 16-bit value to a location in Blackfin memory
Write a 32-bit value to a location in Blackfin memory
Table 1. COMET and DSP memory commands
Page 9 of 13
a
Both tools take a configuration file as input and
produce C code. Lex produces a scanner that
tokenizes incoming strings, and Yacc produces a
program that parses lists of tokens to identify
meaningful patterns (i.e., valid commands). The
Lex/Yacc suite allows for automatic filtering of
incoming strings that do not have commands
associated with them. For general information on
Lex and Yacc, refer to [8].
Yacc Configuration File
[yacc_gmr.txt]
...
%token INTEGER
%token COMET_READ_REG
...
COMET_READ_REG INTEGER INTEGER
{
Lex and Yacc each have a distinct configuration
unsigned char temp;
file. Below are excerpts of each file that show how

to configure the command to read a memorymapped register in one of the COMET framer
chips. The syntax of this command is:
size = sprintf(str,
"COMET_READ_REG 0x%x 0x%x;\n\r", $2, $3);
YY_OUTPUT(str, size);
// echo the command to acknowledge
temp = COMET_readReg($2, $3);
// pass the parameters to a function that
// actually does the COMET access
COMET_READ_REG <base addr> <reg addr>
where
<base addr> is the base address of the COMET
size = sprintf(str,
"COMET_READ_REG (0x%x, 0x%x) -> 0x%x;\n\r",
$2, $3, temp);
chip in the Blackfin
YY_OUTPUT(str, size);
// echo the output
and
<reg addr> is the address of the MMR in the
COMET memory space
Lex Configuration File
[lex_spec.txt]
...
COMET_READ_REG {
return COMET_READ_REG;
// string COMET_READ_REG
returns a distinct token
}
{integer}
yylval = atoi(yytext);
The GNU version of Lex and Yacc were used in

this system. Both Flex, the Lex alternative, and
Bison, the Yacc alternative, are part of the Cygwin
distribution (www.cygwin.com).
The following command line creates the Lex
scanner and Yacc parser.
flex lex_spec.txt [creates lex.yy.c]
bison y d yacc_gmr.txt [creates y.tab.c and y.tab.h]
The resulting C source and header files are

incorporated into the Blackfin VisualDSP++
project.
// yylval is passed to Yacc

return INTEGER;
// any integer returns with this token
}
...
Page 10 of 13
Figure 6. Hyperterminal session
Results
Figure 6 shows a screenshot of the implemented
UART interface protocol. Shown are the initial
status message with the addresses of control
variables and two write_mem commands.
Figure 7 shows a graphical visualization of the
output data in pass-through mode. Inputs are
generated externally (1 kHz sine waves) on a
subset of the 24 T1 channels. ROUT is looped back
externally to SIN, such that all four streams contain
the same data, only delayed in time.
Figure 8 shows an example of how the debugging
flag pins (PF1 and PF3) can be used to show, for
instance, the execution time for the processing of
all channels, or any other useful time in a
relatively straightforward way. The example was
taken from an application where the algorithm
was a line echo canceller (128 ms tail length).
Each of the 32 channels is processed, although
channels 24-31 are not used by a T1 connection.
This was solely intended to demonstrate that the

processing power of the Blackfin processor is
sufficient to do significant amounts of calculations
on all 32 channels, and leaving about 27% of the
total time for other tasks, such as the UART
interface.
Conclusions
This EE-Note, along with schematic drawings and
software framework provides design ideas for a
variety of T1/E1 communication lines.
The main focus is to interface the Blackfin
processors serial ports to standard backplane
formats and to show how to process the PCM
data. This is demonstrated on PCM streams from a
framer device such as the PM4351.
Page 11 of 13
Figure 7. ROUT and SOUT (channel 0) in pass-through mode (RIN is a 1kHz sine-wave for channels 0,1 and 2),
SIN is looped back externally from ROUT.
Figure 8. Execution time (signal PF1, trace 1) and individual channel convergence indicator (signal PF3, trace 4)
in the case of a line echo canceller algorithm. See RESULTS section for more details
Page 12 of 13
a
References
[1] ADSP-BF531/ADSP-BF532/ADSP-BF533 Blackfin Embedded Processor Data Sheet. Rev 0, March 2004.
Analog Devices, Inc.
[2] ADSP-BF533 Blackfin Processor Hardware Reference Manual. Revision 1.0, December 2003.
[3] ADSP-BF561 Blackfin Processor Hardware Reference Manual. Preliminary Revision 0.2, November 2003.
[4] PM4351 COMET Combined E1/T1 Transceiver Datasheet. Issue10, November 2000. PMC-Sierra, Inc.
[5] PM4351 COMET Combined E1/T1/J1 Transceiver/Framer Device Errata. Issue6, April 2002.
PMC-Sierra, Inc.
[6] PM4351 COMET Reference Design. Rev. 2.0, Issue1, November 1998. PMC-Sierra, Inc.
[7] PM4351 COMET Programmers Guide. Issue2, September 2000. PMC-Sierra, Inc.
[8] Lex and Yacc for Embedded Programmers. Embedded Systems Programming.
http://www.embedded.com/story/OEG20030220S0036
Document History
Revision
Description
Rev 1 May 11, 2004

by K. Unterkofler
Initial Release
Page 13 of 13

EE234 V 01

Enviado por

Dados do documento

Título original

Direitos autorais

Formatos disponíveis

Compartilhar este documento

Compartilhar ou incorporar documento

Opções de compartilhamento

Você considera este documento útil?

Este conteúdo é inapropriado?

Direitos autorais:

Formatos disponíveis

EE234 V 01

Enviado por

Direitos autorais:

Formatos disponíveis

Engineer-to-Engineer Note

Interfacing T1/E1 Transceivers/Framers to Blackfin Processors via the

Rev 1 May 11, 2004

[1], [2]. Though it is possible to apply this same

The board was also designed to interface

The software framework is written in C entirely.

Figure 1. System overview

The software also provides a parser-based UART

T1/E1 Data Formats

channels of 8-bit data plus one framing bit

Figure 2. T1/E1 Interface Card Block Diagram

The incoming streams (RIN and SIN) are thus

Figure 3 graphically shows the backplane formats.

manner depicted in the figure as they arrive, that is

Number_of_Samples frames) while the other half is

acquired. This is done in hardware, without any

Figure 3. Backplane signal formats

Note that the hardware companding does not

Similarly, samples going into SIN will reappear

Reference Board Design

1. Check whether a message has been received

The COMET devices are configured via a 9-bit

2. Check whether a new block of received data is

3. Process the data according to the content of

The address decoding scheme adopted in

The reference design, which is configured as a

a. Global pass-through: for initial debugging,

CLKIN, CORECLK, SYSCLK, and SPORT0CLK:

Number_of_Samples: number of frames acquired

Settings for backplane clock source, frame

Assignments of General-Purpose Flags for

DEVICE1, DEVICE2: Base addresses in external

for each block

Number_of_Channels: Set to 32 for a 2.048 MHz

backplane; set to 24 for a true T1 (1.544 MHz)

asynchronous memory of the two COMET

SOFTWARE_COMPANDING: if defined, enabled

companding in software, otherwise the

T1_map: defines settings for T1 (mapping 24

STANDALONE: if this label is commented, the

DUMP_VARIABLES: sends a message with

addresses of important mode variables over

T1 channels onto a 2.048 MHz backplane)

E1, T1: define settings for an E1 backplane @

E1 and T1 COMET configuration

DEBUG_COMET: verifies every write to the

DEBUG_ALGORITHM: uses flag pins to show

execution time of the algorithm, etc. Used for

functions in COMET_init.c according to

Set Exception Handler

Calculate which half of

Detect UART terminal

Press (or send)

Copy Rin to Rout

Initialize COMET devices

Expand Rin and Sin

Loop: For EACH channel

Input Scaling Rin and Sin

Output Scaling Sout

End Channel Loop

Figure 5. Flow Diagram of Main() routine and Processing of PCM samples

echo canceller algorithm, where caching reduced

Using the UART Host Port